globals:
  repo_url: https://github.com/open-mmlab/mmocr/tree/main
  codebase_dir: ../mmocr
  checkpoint_force_download: False
  images:
    img_densetext_det: &img_densetext_det ../mmocr/demo/demo_densetext_det.jpg
    img_demo_text_det: &img_demo_text_det ../mmocr/demo/demo_text_det.jpg
    img_demo_text_recog: &img_demo_text_recog ../mmocr/demo/demo_text_recog.jpg
  metric_info: &metric_info
    hmean-iou: # named after metafile.Results.Metrics
      metric_key: icdar/hmean # eval key name
      tolerance: 0.15 # metric ±n%
    IIIT5K word_acc:
      metric_key: IIIT5K/recog/word_acc_ignore_case_symbol
      tolerance: 0.05 # metric ±n%
      dataset: IIIT5K
    SVT word_acc:
      metric_key: SVT/recog/word_acc_ignore_case_symbol
      tolerance: 0.05 # metric ±n%
      dataset: SVT
    SVTP word_acc:
      metric_key: SVTP/recog/word_acc_ignore_case_symbol
      tolerance: 0.05 # metric ±n%
      dataset: SVTP
  convert_image_det: &convert_image_det
    input_img: *img_densetext_det
    test_img: *img_demo_text_det
  convert_image_rec: &convert_image_rec
    input_img: *img_demo_text_recog
    test_img: *img_demo_text_recog
  backend_test: &default_backend_test True
  sdk:
    sdk_detection_dynamic: &sdk_detection_dynamic configs/mmocr/text-detection/text-detection_sdk_dynamic.py
    sdk_recognition_dynamic: &sdk_recognition_dynamic configs/mmocr/text-recognition/text-recognition_sdk_dynamic.py

onnxruntime:
  # ======= detection =======
  pipeline_ort_detection_static_fp32: &pipeline_ort_detection_static_fp32
    convert_image: *convert_image_det
    deploy_config: configs/mmocr/text-detection/text-detection_onnxruntime_static.py

  pipeline_ort_detection_dynamic_fp32: &pipeline_ort_detection_dynamic_fp32
    convert_image: *convert_image_det
    deploy_config: configs/mmocr/text-detection/text-detection_onnxruntime_dynamic.py

  pipeline_ort_detection_dynamic_fp16: &pipeline_ort_detection_dynamic_fp16
    convert_image: *convert_image_det
    deploy_config: configs/mmocr/text-detection/text-detection_onnxruntime-fp16_dynamic.py

  pipeline_ort_detection_mrcnn_dynamic_fp32: &pipeline_ort_detection_mrcnn_dynamic_fp32
    convert_image: *convert_image_det
    deploy_config: configs/mmocr/text-detection/text-detection_mrcnn_onnxruntime_dynamic.py

  # ======= recognition =======
  pipeline_ort_recognition_static_fp32: &pipeline_ort_recognition_static_fp32
    convert_image: *convert_image_rec
    deploy_config: configs/mmocr/text-recognition/text-recognition_onnxruntime_static.py

  pipeline_ort_recognition_dynamic_fp32: &pipeline_ort_recognition_dynamic_fp32
    convert_image: *convert_image_rec
    deploy_config: configs/mmocr/text-recognition/text-recognition_onnxruntime_dynamic.py

  pipeline_ort_recognition_dynamic_fp16: &pipeline_ort_recognition_dynamic_fp16
    convert_image: *convert_image_rec
    deploy_config: configs/mmocr/text-recognition/text-recognition_onnxruntime-fp16_dynamic.py


tensorrt:
  # ======= detection =======
  pipeline_trt_detection_static_fp32: &pipeline_trt_detection_static_fp32
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt_static-512x512.py

  pipeline_trt_detection_static_fp16: &pipeline_trt_detection_static_fp16
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt-fp16_static-512x512.py

  pipeline_trt_detection_static_int8: &pipeline_trt_detection_static_int8
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt-int8_static-512x512.py

  pipeline_trt_detection_dynamic_fp32: &pipeline_trt_detection_dynamic_fp32
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt_dynamic-320x320-2240x2240.py

  pipeline_trt_detection_mrcnn_dynamic_fp32: &pipeline_trt_detection_mrcnn_dynamic_fp32
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_mrcnn_tensorrt_dynamic-320x320-2240x2240.py

  pipeline_trt_detection_dynamic_fp16: &pipeline_trt_detection_dynamic_fp16
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt-fp16_dynamic-320x320-2240x2240.py

  pipeline_trt_detection_mrcnn_dynamic_fp16: &pipeline_trt_detection_mrcnn_dynamic_fp16
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_mrcnn_tensorrt-fp16_dynamic-320x320-2240x2240.py

  pipeline_trt_detection_dynamic_int8: &pipeline_trt_detection_dynamic_int8
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    sdk_config: *sdk_detection_dynamic
    deploy_config: configs/mmocr/text-detection/text-detection_tensorrt-int8_dynamic-320x320-2240x2240.py

  # ======= recognition =======
  pipeline_trt_recognition_static_fp32_C1: &pipeline_trt_recognition_static_fp32_C1
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt_static-1x32x32.py

  # ABINet models with static shape 32x128
  pipeline_trt_recognition_static_fp32_C3: &pipeline_trt_recognition_static_fp32_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt_static-32x128.py

  pipeline_trt_recognition_static_fp16_C3: &pipeline_trt_recognition_static_fp16_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-fp16_static-32x128.py

  # SAR models with height 48 and channel 3
  pipeline_trt_recognition_dynamic_fp32_H48_C3: &pipeline_trt_recognition_dynamic_fp32_H48_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt_dynamic-48x64-48x640.py

  pipeline_trt_recognition_dynamic_fp16_H48_C3: &pipeline_trt_recognition_dynamic_fp16_H48_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-fp16_dynamic-48x64-48x640.py

  pipeline_trt_recognition_dynamic_int8_H48_C3: &pipeline_trt_recognition_dynamic_int8_H48_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-int8_dynamic-48x64-48x640.py

  # CRNN models with height 32 and channel 1
  pipeline_trt_recognition_dynamic_fp32_H32_C1: &pipeline_trt_recognition_dynamic_fp32_H32_C1
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt_dynamic-1x32x32-1x32x640.py

  pipeline_trt_recognition_dynamic_fp16_H32_C1: &pipeline_trt_recognition_dynamic_fp16_H32_C1
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-fp16_dynamic-1x32x32-1x32x640.py

  pipeline_trt_recognition_dynamic_int8_H32_C1: &pipeline_trt_recognition_dynamic_int8_H32_C1
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-int8_dynamic-1x32x32-1x32x640.py

  # SATRN models with height 32 and channel 3
  pipeline_trt_recognition_dynamic_fp32_H32_C3: &pipeline_trt_recognition_dynamic_fp32_H32_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt_dynamic-32x32-32x640.py

  pipeline_trt_recognition_dynamic_fp16_H32_C3: &pipeline_trt_recognition_dynamic_fp16_H32_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-fp16_dynamic-32x32-32x640.py

  pipeline_trt_recognition_dynamic_int8_H32_C3: &pipeline_trt_recognition_dynamic_int8_H32_C3
    convert_image: *convert_image_rec
    backend_test: *default_backend_test
    sdk_config: *sdk_recognition_dynamic
    deploy_config: configs/mmocr/text-recognition/text-recognition_tensorrt-int8_dynamic-32x32-32x640.py

openvino:
  pipeline_openvino_detection_dynamic_fp32: &pipeline_openvino_detection_dynamic_fp32
    convert_image: *convert_image_det
    backend_test: *default_backend_test
    deploy_config: configs/mmocr/text-detection/text-detection_openvino_dynamic-640x640.py

#  pipeline_openvino_recognition_dynamic_fp32: &pipeline_openvino_recognition_dynamic_fp32
#    convert_image: *convert_image_rec
#    backend_test: *default_backend_test
#    deploy_config:

ncnn:
  pipeline_ncnn_detection_static_fp32: &pipeline_ncnn_detection_static_fp32
    convert_image: *convert_image_det
    backend_test: False
    deploy_config: configs/mmocr/text-detection/text-detection_ncnn_static.py

  pipeline_ncnn_recognition_static_fp32: &pipeline_ncnn_recognition_static_fp32
    convert_image: *convert_image_rec
    backend_test: False
    deploy_config: configs/mmocr/text-recognition/text-recognition_ncnn_static.py

pplnn:
  pipeline_pplnn_detection_dynamic_fp32: &pipeline_pplnn_detection_dynamic_fp32
    convert_image: *convert_image_det
    backend_test: False
    deploy_config: configs/mmocr/text-detection/text-detection_pplnn_dynamic-640x640.py

  pipeline_pplnn_recognition_dynamic_fp32: &pipeline_pplnn_recognition_dynamic_fp32
    convert_image: *convert_image_rec
    backend_test: False
    deploy_config: configs/mmocr/text-recognition/text-recognition_pplnn_dynamic-1x32x32.py

torchscript:
  pipeline_ts_detection_fp32: &pipeline_ts_detection_fp32
    convert_image: *convert_image_det
    backend_test: False
    deploy_config: configs/mmocr/text-detection/text-detection_torchscript.py

  pipeline_ts_detection_mrcnn_fp32: &pipeline_ts_detection_mrcnn_fp32
    convert_image: *convert_image_det
    backend_test: False
    deploy_config: configs/mmocr/text-detection/text-detection_mrcnn_torchscript.py

  pipeline_ts_recognition_fp32: &pipeline_ts_recognition_fp32
    convert_image: *convert_image_rec
    backend_test: False
    deploy_config: configs/mmocr/text-recognition/text-recognition_torchscript.py


models:
  - name: DBNet
    metafile: configs/textdet/dbnet/metafile.yml
    model_configs:
      - configs/textdet/dbnet/dbnet_resnet18_fpnc_1200e_icdar2015.py
    pipelines:
      - *pipeline_ts_detection_fp32
      - *pipeline_ort_detection_dynamic_fp32
      - *pipeline_trt_detection_dynamic_fp16
      - *pipeline_ncnn_detection_static_fp32
      - *pipeline_pplnn_detection_dynamic_fp32
      - *pipeline_openvino_detection_dynamic_fp32

  - name: DBNetpp
    metafile: configs/textdet/dbnetpp/metafile.yml
    model_configs:
      - configs/textdet/dbnetpp/dbnetpp_resnet50_fpnc_1200e_icdar2015.py
    pipelines:
      - *pipeline_ort_detection_dynamic_fp32
      - *pipeline_trt_detection_dynamic_fp16
      - *pipeline_ncnn_detection_static_fp32
      - *pipeline_openvino_detection_dynamic_fp32

  - name: PANet
    metafile: configs/textdet/panet/metafile.yml
    model_configs:
      - configs/textdet/panet/panet_resnet18_fpem-ffm_600e_icdar2015.py
    pipelines:
      - *pipeline_ts_detection_fp32
      - *pipeline_ort_detection_dynamic_fp32
      - *pipeline_trt_detection_dynamic_fp16
      - *pipeline_ncnn_detection_static_fp32
      - *pipeline_pplnn_detection_dynamic_fp32
      - *pipeline_openvino_detection_dynamic_fp32

  - name: PSENet
    metafile: configs/textdet/psenet/metafile.yml
    model_configs:
      - configs/textdet/psenet/psenet_resnet50_fpnf_600e_icdar2015.py
    pipelines:
      - *pipeline_ts_detection_fp32
      - *pipeline_ort_detection_dynamic_fp32
      - *pipeline_trt_detection_dynamic_fp16
      - *pipeline_ncnn_detection_static_fp32
      - *pipeline_pplnn_detection_dynamic_fp32
      - *pipeline_openvino_detection_dynamic_fp32

  - name: TextSnake
    metafile: configs/textdet/textsnake/metafile.yml
    model_configs:
      - configs/textdet/textsnake/textsnake_resnet50_fpn-unet_1200e_ctw1500.py
    pipelines:
      - *pipeline_ts_detection_fp32
      - *pipeline_ort_detection_dynamic_fp32
      - *pipeline_trt_detection_dynamic_fp32

  - name: MaskRCNN
    metafile: configs/textdet/maskrcnn/metafile.yml
    model_configs:
      - configs/textdet/maskrcnn/mask-rcnn_resnet50_fpn_160e_icdar2015.py
    pipelines:
      - *pipeline_ts_detection_mrcnn_fp32
      - *pipeline_ort_detection_mrcnn_dynamic_fp32
      - *pipeline_trt_detection_mrcnn_dynamic_fp32

  - name: CRNN
    metafile: configs/textrecog/crnn/metafile.yml
    model_configs:
      - configs/textrecog/crnn/crnn_mini-vgg_5e_mj.py
    pipelines:
      - *pipeline_ts_recognition_fp32
      - *pipeline_ort_recognition_dynamic_fp32
      - *pipeline_trt_recognition_dynamic_fp16_H32_C1
      - *pipeline_ncnn_recognition_static_fp32
      - *pipeline_pplnn_recognition_dynamic_fp32

  - name: SAR
    metafile: configs/textrecog/sar/metafile.yml
    model_configs:
      - configs/textrecog/sar/sar_resnet31_parallel-decoder_5e_st-sub_mj-sub_sa_real.py
    pipelines:
      - *pipeline_ts_recognition_fp32
      - *pipeline_ort_recognition_dynamic_fp32

  - name: SATRN
    metafile: configs/textrecog/satrn/metafile.yml
    model_configs:
      - configs/textrecog/satrn/satrn_shallow-small_5e_st_mj.py
    pipelines:
      - *pipeline_ts_recognition_fp32
      - *pipeline_ort_recognition_dynamic_fp32
      - *pipeline_trt_recognition_dynamic_fp32_H32_C3

  - name: ABINet
    metafile: configs/textrecog/abinet/metafile.yml
    model_configs:
      - configs/textrecog/abinet/abinet_20e_st-an_mj.py
    pipelines:
      - *pipeline_ts_recognition_fp32
      - *pipeline_ort_recognition_static_fp32
      - *pipeline_trt_recognition_static_fp16_C3